Lack of Effects of Peroxisome Proliferator-Activated Receptor Gamma Genetic Polymorphisms on Breast Cancer Risk : a Case-Control Study and Pooled Analysis

s, 1 Conference Scene, and 1 Meeting Report • •


Introduction
Breast cancer is the most frequently occurring cancer and the most common cause of cancer death among women worldwide; breast cancer accounted for 23% of new cancer cases and 14% of cancer deaths in 2008 (Jemal et al., 2011).It has been suggested that environmental risk factors such as lifestyle, hormonal and reproductive factors, and exposure to chemical carcinogens explain 30-50% of cases; hereditary factors such as high-penetrance susceptibility BRCA 1/2 mutations cause 5-10% of cases; and the other 40-65% can be attributed to unknown factors, such as gene-environment infteractions (Yoo et al., 2006;Park et al., 2009;Yanhua et al., 2012;Mahdi et al., 2013).
Obesity is an alleged risk factor for the development of postmenopausal breast cancer (Calle and Kaaks, 2004;Carmichael and Bates, 2004;Ronco et al., 2012;Sangrajrang et al., 2013).Hypotheses to explain this association include the increased production of estrogen

Lack of Effects of Peroxisome Proliferator-Activated Receptor Gamma Genetic Polymorphisms on Breast Cancer Risk: a Case-Control Study and Pooled Analysis
Boyoung Park 1 , Aesun Shin 2 , Kyee-Zu Kim 2 ,Yeon-Su Lee 3 , Jung-Ah Hwang 3 , Yeonju Kim 4 , Joohon Sung 5 , Keun-Young Yoo 2 , Eun-Sook Lee 6 * in adipose tissue, increased circulating insulin and insulin-like growth factor related to metabolic syndrome, and the tumorigenesis function of adipokines from fat tissue (Lorincz and Sukumar, 2006).Additionally, type 2 diabetes has been suggested to be associated with breast cancer risk (Ronco et al., 2012;Abbastabar et al., 2013).A recent meta-analysis showed that women with type 2 diabetes were 27% more likely to develop breast cancer than other women; even after adjustment for body mass index, a 16% increased risk was still observed (Boyle et al., 2012).These results suggest that the hyperinsulinemia associated with obesity and insulin resistance might be carcinogenic to breast tissue (Minatoya et al., 2013).
Peroxisome proliferator-activated receptor-gamma (PPARγ), a member of the nuclear hormone receptor superfamily, is a transcription factor that plays a major role in lipogenesis, adipogenesis, glucose homeostasis, insulin sensitization, and inflammatory cytokine production (Spiegelman, 1998;He, 2009).PPARγ has been found in various tissues, including endothelial tissue (Kaplan et al., 2007) and normal and malignant epithelium (Mueller et al., 1998), but is expressed at the highest levels in adipose tissues (Auwerx, 1999).Given the role of obesity and insulin insensitivity in breast carcinogenesis, it has been suggested that polymorphisms in the PPARγ gene might be related to the development of breast cancer (Kotta-Loizou et al., 2012).Previous studies showed that PPARγ inhibited the development of preneoplastic lesions in mouse mammary tissue (Mehta et al., 2000), and application of a PPARγ agonist reduced the rate of tumor growth and reversed the malignant phenotype in animal mammary models (Mueller et al., 1998).In human breast carcinoma tissues, a PPARγ activator inhibited the estrogen-mediated proliferation of cancer cells, suggesting PPARγ modulates estrogenic action in human breast cancer cells (Suzuki et al., 2006).In addition, modest efficacy of PPARγ as a chemopreventive target was observed in some cancers, such as prostate cancer and thyroid cancer, and PPARγ agonists inhibited tumor progression in a variety of cancer patients, including breast cancer, colon cancer, or prostate cancer patients, in clinical trials (Grommes et al., 2004;Peters et al., 2012).
The PPARγ polymorphism that has been examined most often in epidemiological studies is PPARγ Pro12 Ala (rs1801282); however, the study populations were restricted to Caucasians, and the results were inconsistent.One study showed an inverse association between PPARγ Pro12 Ala and breast cancer risk (Vogel et al., 2007), while another study showed an increased risk of breast cancer if the patient harbored the PPARγ Pro12 Ala polymorphism (Fratiglioni and Wang, 2007).However, most studies presented no significant association between breast cancer risk and the PPARγ Pro12 Ala polymorphism (Memisoglu et al., 2002;Gallicchio et al., 2007;Justenhoven et al., 2008).A study conducted in East Asia investigated three polymorphisms in the PPARγ gene and suggested that there was no significant effect of the individual PPARγ polymorphisms on breast cancer risk, whereas haplotype analysis showed a significant result (Wu et al., 2011).These discrepancies might be caused by insufficient statistical power due to small sample sizes.Meta-analysis may be a useful way to increase the statistical power by combining the sample sizes of individual studies.
Therefore, we investigated the association between PPARγ genetic polymorphisms and breast cancer risk in an East Asian population and performed a pooled analysis using our results and the results from other studies to analyze the association.

Study participants
The study participants were recruited from the National Cancer Center in Korea.

Genotyping
Six functional single nucleotide polymorphisms (SNPs) in the PPARγ gene (rs4684846, rs1801282, rs2120825, rs2938395, rs1175540, and rs3856806) were selected from previous studies (Paynter et al., 2004;Koh et al., 2006;Vogel et al., 2007).We designed the multiplex PCR and extended primers using the MassARRAY Assay Design software version 3.0 (Sequenom, CA, USA).Genomic DNA was isolated from buffy coats using a DNA Blood Midi M48 Kit (Qiagen, Inc., CA, USA), and 10 ng DNA from each sample was used in the genotyping reaction.The iPLEX Gold assay on the MassARRAY platform (Sequenom), which is based on MALDI-TOF spectrometry, was used for genotyping.The conditions of the PCR and single base extension were the same as previously described (Yoo et al., 2012).To assess the quality of the genotyping assay, 10% of the total samples were run as duplicates, and a concordance test was performed.To check the fitness, genotype clusters were examined manually.Genotype results were collected using the Typer software (Sequenom, version 4.0).

Statistical analysis
The characteristics of the breast cancer cases and controls were compared using a t-test or a chi-square test.The genotype frequencies between the case group and the control group for each of the 2 x 3 contingency tables in the additive model were compared using the Cochrane-Armitage trend test.For the dominant and recessive model, a chi-square test was applied.To estimate   doi.org/10.7314/APJCP.2014.15.21.9093 PPARγ Genetic Polymorphisms and Breast Cancer Risk in Korea the risk of each SNP in the PPARγ gene, multiple logistic regressions were applied, and the results were presented with odds ratio (OR) and 95% confidence intervals (CI).
In the minimally adjusted model, the only variable in this model that was adjusted for was age.In addition to age, the variables with a statistical significance value less than 0.1 in the univariate analysis were adjusted.These included age at menarche, menopause status, hormone replacement therapy, history of being pregnant, number of children, and BMI.We applied dominant, recessive, and codominant models.We used the SAS software version 9.1 (SAS Institute Inc., Cary, NC) for the statistical analyses of the case-control study.

2) Meta-analysis Literature search and data extraction
Literature databases, including PubMed Central and Embase, were searched comprehensively with combinations of the following keywords: "peroxisome proliferator-activated receptor gamma" or "PPARγ", "genotype", "polymorphism", "variant" or "variation", "breast", and "cancer" or "carcinoma".The search was limited to human studies, articles related to breast cancer, and articles written in English.If multiple articles based on the same study population were identified, the study that contained the larger sample size was selected.After removing duplicated articles from the two databases, 15 articles remained.
The studies included in the meta-analysis met all of the following inclusion criteria: (1) independent case-control or cohort studies that evaluated the associations between SNPs in the PPARγ gene and the risk of breast cancer; (2) sufficient data for the calculation of crude OR with CI; (3) breast cancer cases regardless of stage, hormone receptor status, menopausal status, and histological type; and (4) articles written in English.Exclusion criteria included the following: (1) not an original article (n=5); (2) non-breast cancer patients (n=1); (3) studies regarding PPARγ coactivators (n=2); and (4) a duplicated population of the current case-control study (n=1) (Kim et al., 2012).Additionally, the results from this case-control study were included.Therefore, two articles including this study were analyzed for rs3856806, and six articles including this study were analyzed for rs1801282 (Figure 1).The following information was extracted from each study: the first author's name, year of publication, country, ethnicity, source of controls, menopausal status, and the number of subjects in each genotype in the cases and controls.Two authors independently assessed the articles for compliance with the inclusion/exclusion criteria and reached a consistent decision.

Statistical analysis
The associations between polymorphisms in the PPARγ gene and the risk of breast cancer were assessed by calculating the pooled OR and 95% CI.Associations under three different types of ORs were calculated using the codominant model, the dominant model, and the recessive model.Q statistics were used to investigate the heterogeneity between studies.A p-value greater than 0.05 indicated that there was no significant heterogeneity, allowing for the applicability of the fixed effects model (Mantel-Haenszel method) (Mantel and Haenszel, 1959).A Begg's funnel plot was generated for rs1801282 to detect bias or systematic heterogeneity (Begg and Mazumdar, 1994), and an Egger's test was used to estimate the publication bias.A p-value of 0.05 or lower was considered a statistically significant publication bias (Egger et al., 1997).Sensitivity analysis was performed by excluding each study in turn for the rs1801282 polymorphism.All statistical tests in the meta-analysis were performed using the STATA software version 12.0 (StataCorp LP, College Station, TX, USA).

1) Case-control study
The basic characteristics of the study participants are shown in Table 1.The patient's age at menarche, the proportion of nullipara, and the proportion of participants with a BMI <25 kg/m2 were significantly different between the cases and controls, with a p-value ≤ 0.05.
The genotype distributions of the 6 SNPs in the PPARγ gene between the cases and the controls are presented in Table 2. None of the 6 PPARγ polymorphisms had a significantly different distribution between the cases and the controls (p-values range from 0.29-0.90).Both in the codominant and the recessive models, the TT genotype of rs3856806 was related to a decreased risk of breast cancer with marginal significance after adjustment (OR: 0.43, 95% CI: 0.17-1.10compared to the CC genotype; OR: 0.42, 95% CI: 0.16-1.07compared to the CC+CT genotype).No significant associations were observed with any of the other five SNPs (rs4684846, rs1801282, rs2120825, rs2938395, and rs1175540).

2) Meta-analysis results
The characteristics of the studies included in the metaanalysis are summarized in Table 3.All studies regarding the rs3856806 SNP were hospital-based case-control studies and included East Asians.Among the six studies that investigated rs1801282, five studies included mostly Caucasians and were all population-based case-control studies.Only one study that examined the rs1801282 SNP and breast cancer risk included East Asians, and this study was a hospital-based case-control study.
The results of the meta-analysis including the current case-control study are presented in Table 4. Significant heterogeneities were not found between the studies, according to the Q statistic results.Therefore, a fixed effects model was employed.When examining the  28.9) 0.11 Body mass index (<25.0kg/m2) (N, %) 305(66.9)353 (76.6) <0.01 rs3856806 polymorphism, the results of the analysis including 2 studies consisting of 747 cases and 1,030 controls showed that there was no significant association between the rs3856806 polymorphism and breast cancer risk.For the rs1801282 polymorphism, 6 studies with 2,668 cases and 3,764 controls were examined, and the results suggested that there was no significant association between the rs1801282 polymorphism and breast cancer risk (CG vs. CC: OR: 0.91, 95% CI: 0.79-1.04;CG+GG vs. CC: OR: 0.92, 95% CI: 0.81-1.05).These pooled estimates for all the genetic models were insensitive to the exclusion of individual studies, demonstrating the statistical robustness of the results (data not shown).No publication bias was observed according to the Begg's

Discussion
This hospital based case-control study and metaanalysis suggest there is no association between PPARγ polymorphisms and breast cancer risk.We examined the association between the rs1801282 polymorphism, which is the most studied PPARγ polymorphism, and breast cancer risk for the first time in Asian females in a case-control study.Additionally, this is the first systematic review and pooled meta-analysis regarding PPARγ polymorphisms, although several previous meta-analyses were conducted for other adiposity related genes, such as leptin or leptin receptor polymorphisms (Liu and Liu, 2011;Wang et al., 2012).
Obesity and type 2 diabetes are important risk factors for breast cancer and are correlated with a worse prognosis (Carmichael and Bates, 2004;Barone et al., 2008).These two risk factors share biological mechanisms that affect breast cancer tissue.These mechanisms include the direct effect of insulin on the proliferation of breast cancer cells, increased estrogen production and bioavailability, and changes in adipokines (Vona-Davis and Rose, 2012).The functional variants of the PPARγ gene are related to both lipid metabolism and insulin sensitivity and can affect obesity, type 2 diabetes, and diabetic complications; these effects have been confirmed by several epidemiological studies, including meta-analysis (Masud and Ye, 2003;Gouda et al., 2010).Although the biological functions of PPARγ in cancer development, such as promoting terminal differentiation, inhibiting cell growth, increasing apoptosis in human cancer cell lines, and inhibiting tumorigenesis (Peters et al., 2012), have been well documented, epidemiological studies regarding the associations between PPARγ polymorphisms and breast cancer risk, including the case-control study presented here, have shown inconclusive results with relatively small sample sizes.In this pooled analysis that combined the sample sizes of individual epidemiological studies, we did not find a significant association between two PPARγ polymorphisms (rs3856806 and rs1801282) and breast cancer risk.
Several limitations of this study should be mentioned.In this case-control study, we obtained information on the adjusted factors, such as menstrual and reproductive factors, from self-administered questionnaires, and the possibility of information bias cannot be ruled out.However, this is a non-differential misclassification, and the effects of misclassification on the results are expected to be minimal.Because correction for multiple comparisons was not performed in this study, the estimates should be interpreted cautiously.Additionally, we conducted a candidate gene association study and did not consider linkage disequilibrium or the function of combined SNPs, as previous studies have done (Li et al., 2011;Wu et al., 2011).Considering the linkage disequilibrium across the population, the selected SNPs in this study were positional candidates, rather than functional candidates.
There are also several limitations to the meta-analysis.First, the control populations were not uniformly defined.Although the study populations of the two studies included in the meta-analysis of rs3856806 were East Asian populations, there might be differences in the genetic distribution, gene effects, or gene-environmental interactions between countries.Additionally, for metaanalysis of rs1801282, Caucasians, East Asians, and other ethnicities were mixed.Although we did not find different results in the sensitivity analysis, there might be different effects of ethnicity on the genetic predisposition to human diseases, as many previous studies have shown (Pan et al., 2005).For the rs1801282 polymorphism, although most studies selected controls from healthy populations, one study used participants with benign breast diseases as controls (Gallicchio et al., 2007), and this case-control study used participants in a cancer screening program that did not have abnormal findings during the screening.Therefore, non-differential misclassification was possible because those with benign breast diseases may have risks of developing breast cancer.Second, because we could not confirm the menopausal status or BMI of the study populations included in the meta-analysis, this meta-analysis was unable to address gene-environmental interactions that could be important factors in the association of PPARγ polymorphisms and breast cancer risk.Additionally, the pooled analysis was performed only on the basis of the number of patients with each type of polymorphism, and unadjusted estimates were calculated.Therefore, a more precise analysis should be conducted if the confounding factors of the individuals are available.Third, because there were only two studies for the rs3856806 polymorphism, we could not assess the publication bias or perform sensitivity analysis.Fourth, although we combined all available data from the literature, the number of populations included in the pooled analysis was not enough to obtain a high statistical power.Fifth, in the meta-analysis, we did not perform corrections for multiple comparisons.
In conclusion, this case-control study and metaanalysis suggest there is no significant association between PPARγ polymorphisms and the risk of developing breast cancer despite the biological effects that PPARγ has on

Figure 1 .
Figure 1.Flow Chart of Meta-analysis for Exclusion/ Inclusion of the Studies.*Including 1 review Article, 2 Abstracts, 1 Conference Scene, and 1 Meeting Report

Table 3 . Characteristics of the Studies Included in the Meta-analysis
HB, hospital-based; PB, population-based.

Table 2 . PPARγ Genotype Frequency and Odds Ratios for Breast Cancer Risk, according to PPARγ Genetic Polymorphisms in a Case-control Study among Korean Women
Cochrane-Armitage test for trend was applied for additive model and chi-square test was applied for dominant and recessive model; 2 Adjusted for age; 3 Adjusted for age, age at menarche, menopause status, hormone replacement therapy, pregnancy, number of children, and body mass index DOI:http://dx.doi.org/10.7314/APJCP.2014.15.21.9093PPARγ Genetic Polymorphisms and Breast Cancer Risk in Korea funnel plot and the Egger's test, which had a p-value of 0.21.

Table 4 . Summary Odd Ratios and 95% Confidence In- tervals of the Association between the PPARγ Polymor- phisms and Breast Cancer Risk Using a Fixed Model
Boyoung Park et al Asian Pacific Journal of Cancer Prevention, Vol 15, 2014 9098 breast carcinogenesis.